ABSTRACT
Immunization against a 70 kDa band that co-purifies with skeletal muscle integrins has resulted in an anti-body directed against the avian 7 integrin subunit. The specificity of the antibody was established by patterns of tissue staining and cross-reactivity with antibodies directed against the cytoplasmic domain of the rat 7 cytoplasmic domain. On sections of adult skeletal muscle the 7 integrin was enriched in the myotendinous junc-tion (MTJ). This localization was unique as neither the 1, 3, 5, 6 and v subunit localizes in the myotendi-nous junction. The distribution of the 7 subunit in the MTJ was examined during embryonic development. 7 expression in the junction is first apparent around embryo day 14 and is almost exclusively at the devel-oping MTJ at this stage. 3 is expressed with distinctive punctate staining around the junctional area in earlier embryos (11-day). The time of appearance of the 7 sub-unit in the MTJ correlates with the insertion of myofib-rils into subsarcolemmal densities and folding of the junctional membrane, suggesting a role of the 7 inte-grin in this process. Vinculin is present throughout development of the myotendinous junction, suggesting that the 7 integrin recognizes a preformed cytoskele-tal structure. The presence of the 7 subunit in the myotendinous junction and the 5 subunit in the adhe-sion plaque demonstrates a molecular difference between these two adherens junctions. It also points to possible origins of junctional specificity on muscle. Dif-ferences between these two junctions were developed further using an antibody against phosphotyrosine (PY20). Phosphotyrosine is thought to participate in the organization and stabilization of adhesions. The focal adhesion and the neuromuscular junction, but not the MTJ, contained proteins phosphorylated on tyrosine.
INTRODUCTION
The β1 integrins are a family of cell surface molecules that mediate interactions between the extracellular matrix and cytoskeleton (reviewed by Buck and Horwitz, 1987; Hynes, 1987, 1992). Accumulating evidence has demonstrated that integrins play a central role in muscle development. A mon-oclonal antibody against the β1 subunit inhibits adhesion and perturbs cytoskeletal integrity, is localized in all major developing junctional regions as well as along much of the sarcolemmal surface, and reversibly inhibits terminal myo-genic differentiation (Neff et al., 1982; Bozyczko et al., 1989; Menko and Boettiger, 1987) The importance of inte-grin in myogenesis and sarcomeric cytoarchitecture forma-tion is further demonstrated by the study of the Drosophila lethal myospheroid mutant. This mutant lacks β1 integrin and displays grossly abnormal muscle morphology due largely to detachment at the tendon region and improper myofibril organization (Newman and Wright, 1981; Volk et al., 1990).
In adult muscle, the interactions with extracellular matrix occur along the sarcolemma adjacent to the basal lamina and in specialized, morphologically distinct, junctional areas such as the myotendinous and neuromuscular junc-tions and the costamere. The myotendinous junction (MTJ), the focus of the present study, is the principal structure that transmits force generated by muscle contraction to the tendon. Prominent structural features of the MTJ observed by electron microscopy include extensive invagination of the sarcolemma at the interface between muscle and tendon at sites of myofibrillar termination (Mackay et al., 1969; Trotter et al., 1981). These invaginations serve to increase the area of muscle-tendon contact, thus decreasing the con-tractile stress applied to the junctional sarcolemma (Tidball, 1983; Trotter et al., 1985).
At least nine α subunits, α1 through α8 and αv, associ-ate with the β1 integrin subunit (Hynes, 1992). These α subunits play a major role in ligand binding specificity. Recently, the α5 subunit has been localized in focal adhe-sion-like structures, but not in the myotendinous junction, on myogenic cultures (Lakonishok et al., 1992). This speci-ficity raises the possibility that differences among the junc-tions might arise from the different α subunits that popu-late them. The presence of such molecular differences would serve not only to distinguish among junctions on muscle, but also to provide clues to the mechanism of their organization and development. These studies were recently made feasibile by the availability of an increasing number of antibodies that react with chicken α subunits (Syfrig et al., 1991; Hynes et al., 1989; Muschler and Horwitz, 1991; Bronner-Fraser et al., 1992; Bossy and Reichardt, 1990).
Here, we report characterization of a monoclonal anti-body against the chicken integrin α7 subunit. Using this antibody and antibodies against the other α subunits, we have found that the α7 subunit is the major integrin present in the adult skeletal muscle myotendinous junction. This contrasts the muscle adhesion plaque in which the α5 inte-grin is the major integrin present.
MATERIALS AND METHODS
Affinity purification of the 1 integrins
The β1 integrins were purified from adult chicken tissues (Pel-Freez, Rogers, AR) on a CSAT affinity column conjugated with the monoclonal antibody against the β1 subunit (Bozyczko et al., 1989). The tissues were homogenized in a Waring blendor at 4°C in 4 volumes of extraction buffer (10 mM Tris, 150 mM NaCl,
0.5 mM CaCl 2, 1.0 μg/ml pepstatin, 1.0 μg/ml leupeptin, 1.0 mM 1,10-phenanthroline, 0.1 mM PMSF, pH 8.0). The homogenate was then centrifuged at 25,000 g for 20 minutes, and the super-natant decanted. Acetic acid was added to the supernatant to a final concentration of 20 mM. After stirring at 4°C for 45 min-utes, the extract was centrifuged again at 25,000 g for 20 min-utes. to remove the precipitate. The supernatant was neutralized with 1.0 M Tris-base and then passed over a CSAT mAb affin-ity column. The column was washed extensively with extraction buffer. The proteins bound to the column were eluted in extrac-tion buffer containing 50 mM diethylamine at pH 11.5. Purifica-tion of the G9 antigens was performed in an identical way to that described for the CSAT antigen.
Monoclonal and polyclonal antibodies
The H1 monoclonal antibody against the 70 kDa protein was obtained by immunizing 9-week-old female Balb/c mice at 21-day intervals with approximately 5-10 μg purified protein. The immunogen was prepared from the integrins purified from chicken cardiac muscle by the CSAT column. The integrins were separ-ated on SDS-PAGE, and the protein bands were visualized with Coomassie blue stain. The 70 kDa band was excised and elec-troeluted using a Centrilutor from Amicon (Beverly, MA). The electroeluted proteins were then dialyzed, concentrated, and used for immunization. Two weeks after the third immunization the sera from the mice was positive for the 70 kDa protein on immunoblots. A fusion was performed by the method of Kennett et al. (1982). Among the monoclonal antibodies against the 70 kDa obtained, H1 and G9 were characterized in more detail. The H1 mAb does not recognize native proteins and can only purify its antigen when it is denatured. H1 immunoblots and immuno-fluorescent stains well. G9 mAb purifies the 70 kDa protein that is recognized by H1 in addition to the β1, α6 and possibly α3, but not α5 subunits. The G9 mAb is used primarily to purify the H1 antigen as it enriches and gives a higher yield of the 70 kDa band than does the CSAT mAb.
Monoclonal antibodies against chicken integrin β1 (CSAT, W1B10), α5 (A2) and α6 subunit (C6) have been described pre-viously (Neff et al., 1982; Hayashi et al., 1990; Muschler and Hor-witz, 1991; Bronner-Fraser et al., 1992). Polyclonal antisera against chicken integrin α1 was kindly provided by Drs J. Syfrig and M. Paulsson (Syfrig et al., 1991). Rabbit antisera raised against polypeptides modeled after chicken α3 cytoplasmic domain (RIQPSETERLTDDY) (Hynes et al., 1989) and chicken αv cytoplasmic domain (CKRVRPPQEEQEREQLQPHENGEGT-SEA) (Bossy and Reichardt, 1990), were gifts from C. Buck and L. Reichardt, respectively. Polyclonal antisera against a polypep-tide from the rat α7 cytoplasmic domain (EDRQQFKEEKTG-TIQRSNWGNSQWEGS) were kindly provided by S. Kaufman. Monoclonal antibodies against phosphotyrosine (PY20) and vin-culin (VIN-11-5) were purchased from Sigma (St. Louis, MO). A rabbit antibody against chicken dystrophin is a gift from F. Pons. Hybridoma supernatant against talin (8E6) and tenascin (M1-B4) were obtained from Developmental Studies Hybridoma Bank (Johns Hopkins University, School of Medicine, Baltimore, MD and University of Iowa, Iowa City, IA). The P3 mAb, derived from P3/X63-Ag8, was used as a negative control.
Immunoblotting
Proteins were separated by SDS-PAGE (Laemmli, 1970) and transferred to nitrocellulose paper (Schleicher & Schuell, Keene, NH). The nitrocellulose paper was then blocked in 3% nonfat dry milk in TS buffer (10 mM Tris-HCl, 150 mM NaCl, pH 7.4) for 1 hour. The blots were developed by using the VectaStain alka-line phosphatase kit (Vector Labs, Inc., Burlingame, CA). The pri-mary antibodies were used at a concentration of 10-20 μg/ml. Immunoblotting using the rat α7 cytodomain antibody (αCD) was done similarly, except by using 4% gelatin in TBST buffer (10 mM Tris-HCl, 150 mM NaCl and 0.05% Tween-20, pH 8.0) for blocking the nitrocellulose and 2% gelatin in TBST for all anti-body dilution and washes.
Cryosection and immunofluorescent staining
Embryonic and adult chicken hind-leg muscle tissues were dis-sected from white leghorn chickens obtained from the University of Illinois poultry farms. The tissues were flash frozen in OCT compound (Miles Inc., Elkhart, IN) by emersion in liquid nitrogen. Sections (8-10 μm) were cut at −20°C in a Tissue Tek cryostat and placed on glass slides coated with 0.5% gelatin.
For immunofluorescence staining, chicken embryo fibroblasts were cultured in Dulbecco’s Modified Eagle’s medium, supple-mented with 5% fetal calf serum on glass coverslips coated with 10 μg/ml fibronectin for 24 hours. Two hours prior to staining, the cells were transferred into medium without serum.
The cryosections or cultured cells were fixed with 2% formaldehyde (Sigma) in PBS for 15 minutes, then quenched with 0.15 M glycine. The samples were blocked in 5% goat serum in PBS for 30 minutes. After blocking, the samples were incubated for 30 minutes: first with primary antibody and then with the appropriate species-specific FITC-or rhodamine-con-jugated secondary antibody (Fab2 fragments, Cappel Labs, Malvern, PA) diluted into the blocking solution. After each anti-body application the samples were washed with PBS extensively. Hybridoma supernatant was used either undiluted or diluted 1:1. Purified antibody was used at concentration of 10-20 μg/ml. For immunofluorescent staining of tissue culture cells, the cells were permeablized after fixation with 0.3% Triton X-100 for 5 min-utes.
Acetylcholine receptor clusters were visualized by incubating sections with 10 μg/ml rhodamine-conjugated α-bungarotoxin (Molecular Probes, Eugene, OR) for 45 minutes.
Immunoprecipitation
Since H1 mAb does not readily react with the native protein, it was necessary to denature the proteins in order to immunopre-cipitate its antigen. The β1 integrins purified from the G9 column were denatured in 6 M guanidine isothiocyanate by heating to 90°C for 2 minutes. They were then dialyzed and incubated with H1 mAb conjugated to Sepharose beads (Pharmacia, Piscataway, NJ). The beads were washed 3 × in extraction buffer, and then 3 × in extraction buffer including 0.5 M MgCl2. The bound proteins were eluted in SDS-PAGE sample buffer.
RESULTS
Characterization of the 7H1 monoclonal antibody
The presence of a 70 kDa protein band was reported pre-viously in purifications of β1 integrins from cardiac or skeletal muscle on a CSAT affinity column (Bozyczko et al., 1989) (Fig. 1A, lane 1). Monoclonal antibodies against the 70 kDa protein were generated by immunizing mice with the protein purified by electroelution. As they were raised against denatured protein, most blotted the 70 kDa band but were not useful for purification. H1, an example of this class, was characterized further. One mAb, G9, did not blot; but it did purify the 70 kDa protein along with the β1, α6 and (possibly) α3 subunits, but not the prominent α5 subunit (Fig. 1A, lane 2). The reason for this complex purification profile is unclear and could arise either from a common epitope on the α subunits or a common binding molecule. In any event it provided a useful procedure for partially purifying and enriching the 70 kDa band.
Immunofluorescent staining of adult tissue sections using the H1 monoclonal antibody demonstrated that the antigen is localized only on limited types of tissues: skeletal muscle, cardiac muscle and blood vessels, but not on gizzard smooth muscle, brain and sciatic nerve tissues (data not shown). The H1 mAb immunoblotted a strong band at 70 kDa and two weak bands around 120 kDa and 96 kDa (Fig. 1B, lane 1) on integrins purified by the G9 column. The tissue dis-tribution and multiple-band pattern seen by immunoblotting are very similar to those of the rat α7 integrin subunit reported by Song et al. (1992). The rat α7 subunit has bands at 120, 100 and 70 kDa, which result from proteolytic cleav-age sites in the rat α7 subunit. Although protease inhibitors were included in all solutions during the purification pro-cedure, the 70 kDa fragment was always the major band recognized by the H1 mAb.
Cross-reactivity between the α7 integrin and the H1 anti-gen was shown using an antibody that recognizes the cyto-plasmic domain of the rat α7 subunit. Fig. 1B (lane 2) shows that this anti-α7 cytoplasmic domain antibody immunoblots a strong band at 70 kDa and two weak bands around 120 and 96 kDa on the chicken integrins purified by the G9 column. This blotting is indistinguishable from that of the H1 mAb (Fig.1B, lane 1). The immunoblotting is inhibited by addition of the peptide against which the antiserum was raised (Fig. 1B, lane 3). Furthermore, the antibody raised against the rat α7 cytoplasmic domain immunoblotted the 70 kDa band purified by H1 mAb following denaturation and dissociation into monomers (Fig. 2). Thus the immuno-logical relation, similar tissue distribution and protease degradation pattern, argue strongly that the H1 antigen is the chicken homolog of the rat α7. Therefore, this antigen will be referred to as the chicken α7 subunit and the H1 antibody as α7H1.
The α7 subunit is enriched in the myotendinous junction on adult skeletal muscle tissue
As skeletal muscle is the major tissue in which α7 is expressed, we examined its localization further using adult hind-leg muscle. As shown in Fig. 3B, α7H1 stains weakly along the sarcolemma of the muscle cells. However, stain-ing in the region where the muscle cell terminates at the tendon (the wavy-looking structure in the phase-contrast image) (Fig. 3A) was bright. Both vinculin (Fig. 3C) and talin (not shown) are highly enriched in this area, confirm-ing its identity as a myotendinous junction (Shear and Bloch, 1985; Tidball et al., 1986). The staining pattern by α7H1 mAb on the muscle tip is essentially identical to that of vinculin: both show intense ‘capping’ at the junction. A similar enrichment is also seen on adult rat skeletal muscle (S. Kaufman and M. George-Weinstein, personal commu-nication).
The localization of α7 in the MTJ appears to be unique among the β1 integrins assayed. An antibody against the chicken α1 subunit stains on the plasma membrane of the muscle but is not detected in the myotendinous junction (Fig. 4A). Furthermore, antibodies against the α3, α5, α6 and αv subunits also do not stain the junction (Fig. 4B-E). Some weak staining in the region of the myotendinous junc-tion was observed with α5 and αv antibodies. However, their staining lacks the typical enrichment at the myotube tip. Therefore, it is more likely to be either sarcolemmal or connective tissue, rather than junctional, staining. As expected, the staining of β1 subunit is highly enriched in the junctional area in addition to its strong staining on the other parts of muscle plasma membrane (Fig. 4F).
Expression of integrin subunits during development of the MTJ
The changes in the interaction between muscle and tendon and the formation of myotendinous junctions during devel-opment have been addressed previously only at the mor-phological level (Tidball and Lin, 1989). Therefore, we investigated the distribution of integrin α subunits at regions of muscle-tendon interaction during embryonic development. We used 11-, 14-, 18- and 20-day-old embryo hind-leg cross-sections, as this is the developmental window during which myotendinous junctions form. As shown in Fig. 5, the integrin α7 subunit is present in the MTJ area in 14-, 18- and 20-day-old embryos. However, the appearance of α7 integrin changes during this time. The α7 staining in the myotendinous junction is considerably brighter in the older embryos. While very weak staining of α7 on the rest of muscle plasma membrane can be seen in the 20-day-old embryos, the α7 expression can be detected only in the junctional area in the earlier embryo. Further-more, the staining of α7H1 in the 14-day embryo (Fig. 5e) appears more discontinuous, and the junctional localization is not as well defined as it is in the 20-day-old embryo, a stage at which the staining appears concentrated at the muscle plasma membrane that contacts tendon.
We have also examined even younger embryos to deter-mine the molecular components in the regions of early muscle tendon interaction. In the 11-day-old embryo, the tendon could not be identified with certainty from the phase-contrast image. Therefore, an antibody against tenascin (Fig. 6a) and collagen I (not shown) was used along with the morphological criteria to mark the tendon region specifically. Antibodies against dystrophin and desmin (not shown) identified the muscle region (Fig. 6b). Vinculin stains at the boundary region of muscle-tendon contact (Fig. 6c), whereas α7 staining is not apparent (Fig. 6f). However, an antibody against α3 stained positively in the junctional area (Fig. 6d). The staining of α3 is enriched at but not restricted to the junctional region, and while it is in the vicinity of vinculin staining, it did not co-localize with vinculin. (Fig. 6c and d). In addition, the staining pat-tern of α3 is distinctively punctate. Interestingly, we have not been able to detect the the α7 integrin at myotube tips in skeletal muscle cultures even as old as 21 days. How-ever, the α3 subunit, while not localized specifically to the myotube tip, often appears to co-localize with myofibrils along portions of their length including their ends.
The absence of proteins phosphorylated on tyrosine distinguishes the myotendinous junction from the focal adhesion
As a type of adherens junction, the myotendinous junction shares many molecular, structural and functional charac-teristics with focal adhesions in fibroblasts (Burridge et al., 1988). Our observation that the α7 but not the α5 integrin localizes specifically to this junction points to molecular differences between these two junctions. Whereas previous studies have demonstrated several molecular similarities between the focal adhesion and myotendinous junction, the difference in amino acid sequence among the cytoplasmic domains of the different integrin α subunits points to poten-tially different integrin-associated cytoskeletal molecules in these junctions.
Recent interest in the focal adhesions has focussed on the enrichment in focal adhesions of several proteins that are phosphorylated on tyrosine residues. These can be detected with a monoclonal antibody PY20 (Burridge et al., 1992). This class of molecules is particularly interesting as they are thought to be involved in signalling as well as organization and stabilization of adhesions. We assayed for the presence of such molecules in the myotendinous junc-tion using the antibody PY20. Surprisingly, there is no detectable staining in the in MTJ (Fig. 7A). In contrast, phosphotyrosine staining is seen in both the fibroblast adhesion plaque (Fig. 7B) and the neuromuscular junction (Fig. 7C).
DISCUSSION
In this paper, we report the characterization of a monoclonal antibody, α7H1, directed against the chicken integrin α7 subunit. As muscle is the major tissue on which α7 inte-grin is expressed, we investigated its localization on skele-tal muscle using this antibody. The study was prompted by previous observations that the β1 subunit occupies several specialized regions on skeletal muscle including the myotendinous and neuromuscular junctions (Bozyczko et al., 1989). Furthermore, it was reported recently that α5 integrin localizes specifically in the focal adhesion-like structure on skeletal muscle, thus raising the possibility that integrin α subunits might distinguish between sarcoplasma specializations (Lakonishok et al., 1992). We now find that the α7 subunit is highly enriched in the myotendinous junc-tion in addition to its weaker presence along other parts of the membrane. The selective presence of the α7 subunit in the MTJ points to the α7 integrin as a determinant of junc-tional specificity, serves to distinguish further the MTJ from other adherens junctions, and provides a molecular marker of the MTJ during development.
Like several other integrin α subunits (Hynes et al., 1989; Muschler and Horwitz, 1991; Bronner-Fraser et al., 1992; Bossy and Reichardt, 1990), the cytoplasmic domain of α7 appears to be conserved across species. Rabbit polyclonal antisera raised against the cytoplasmic domain of rat α7 rec-ognizes the chicken protein. In addition to an immunolog-ical relation, the chicken α7 subunit shares other charac-teristics with the rat α7 subunit, including a selective tissue distribution and a characteristic protease degradation pat-tern. Two other integrin subunits are reported to undergo regulated protease cleavage. The β4 integrin undergoes tissue-specific proteolytic processing (Giancotti et al., 1992), and the cleavage of the α4 integrin appears to cor-relate with T lymphocyte activation (Wayner et al., 1989). The physiological significance of the α7 cleavage is not clear; however, its prominent presence even in muscle extracts prepared rapidly with protease inhibitors suggests that it may be regulated and important rather than a degradative cleavage.
The localization of the integrin α7 subunit in the MTJ appears specific. Other α subunits that are expressed on adult muscle cell, e.g. α1, α3, α6 and αv, do not localize in the MTJ, but instead are found on other regions of the plasma membrane. The specificity may lie in the ECM ligand for α7, which is believed to be laminin (Kramer et al., 1991; Song et al., 1992). However, laminin is present ubiquitously in the basement membrane area around the muscle cell as well as in the myotendinous junction (Swas-dison and Mayne, 1989), whereas the α7 integrin is highly concentrated in the junctional region..
Alternatively, the interaction of the α7 integrin with cytoskeletal components may direct α7 localization in the MTJ. Several reports have described interactions between the integrin β1 subunit cytoplasmic domain and cytoskele-ton associated molecules (Horwitz et al., 1986; Solowska et al., 1989; Hayashi et al., 1990; Otey et al., 1990; LaFlamme et al., 1992). There is still no direct evidence demonstrating interactions between α subunit cytoplasmic domains and cytoskeleton-associated proteins. However, a role for the α subunit in determining localization to focal adhesions has been demonstrated (LaFlamme et al., 1992). Furthermore, the very different cytoplasmic domain sequences of the α subunits that are highly conserved among species also argues for a specificity determining role. The α7 cytoplasmic domain is particularly interesting in this regard as it is also relatively large, with 77 amino acids (Song et al., 1992).
As an adherens junction, the myotendinous junction shares structural and functional characteristics with focal adhesion on fibroblasts. Both are specialized regions of the plasma membrane that anchor actin filaments and transmit tension across the membrane to the extracellular matrix. The shear stress placed on the MTJ is similar in type and magnitude to those occurring at the focal adhesions (Tid-ball et al., 1986). In addition, several cytoplasmic proteins in focal adhesion have also been identified in the MTJ including vinculin (Shear and Bloch, 1985), talin (Tidball et al., 1986), paxillin (Turner et al., 1991) and dystrophin (Kramarcy and Sealock, 1990; Masuda et al., 1992). Despite these similarities, our results show that the integrin α subunits in the MTJ and focal adhesion are different. α5and αv are found in focal adhesions in many types of cells, but are absent from MTJ.
We have been able to differentiate these two structures further. Recently, several proteins phosphorylated on tyro-sine have been found in focal adhesions. These include pp125FAK (Schaller et al., 1992), tensin (Davis et al., 1991) and paxillin (Burridge et al., 1992). The presence of these phosphoproteins is readily visualized using a monoclonal antibody raised against phosphotyrosine (PY20), which stains the focal adhesion (Burridge et al., 1992). However, our results indicate that proteins with phosphorylated tyrosines are not prominent in the adult MTJ; but, in con-trast, are present in neuromuscular junctions. The neuro-muscular junction staining is consistent with the finding by Qu et al. (1990) that junctional acetylcholine receptors are phosphorylated on tyrosine in the rat diaphragm. Other mol-ecular differences between the cytoskeleton-associated mol-ecules in the myotendinous junction and the adhesion plaque have been reported. α-Actinin is in focal adhesions but is absent from myotendinous junctions (Tidball, 1987).
The development of myotendinous junctions has been described previously only at the morphological level. By light and electron microscopy, Tidball and Lin (1989) have described the time course of the major events in MTJ for-mation. In the hind leg of 11-day chicken embryos, myofib-rils are only infrequently inserted into the ends of myotubes, and the region of apposition between fibroblasts and myotubes can be distinguised in the light microscope. By embryo day 15, myofibrils insert into subsarcolemmal den-sities at incipient MTJs, and folding of the junctional membrane can be observed for the first time. Our results extend these observations to the molecular level. We have shown that vinculin is a very early marker for the myotendinous junction and remains there throughout its development. The α7 appears later than vinculin, suggesting that its localiza-tion is directed by a nascent, partially organized cytoskele-ton. α7 first appears in 14-day embryos. Its presence cor-relates with a structural change in the nascent myotendinous junction, i.e. the insertion of the myofibrils into the sub-sarcolemmal densities. This suggests that α7 may play a role in the very early organization of the junction includ-ing membrane folding and myofibril insertion. The α3 sub-unit is seen in the 11-day embryo; however, its staining is very punctate and does not apppear to colocalize with vin-culin. The role of α3 enrichment at the boundary region remains to be explored but may function to organize or sta-bilize the myofibrils prior to insertion at the tendon.
ACKNOWLEDGEMENTS
The authors thank J. Syfrig (M. E. Müller -Institute for Bio-mechanics, Bern, Switzerland), M. Paulsson (University of Basel, Basel, Switzerland), C. Buck (Wistar Institute, Philadelphia, PA), L. Reichardt (University of California at San Francisco, San Fran-cisco, CA), K. Burridge (University of North Carolina, Chapel Hill) and F. Pons (Inserm, Montpellier, France) for generously providing antibodies used in this study. This work was supported by NIH grant GM 23244.